Abstract: This dissertation brings forward the method of automatic abstracting’s realization based on text clustering and natural language understanding. This method can overcome the shortage of the automatic abstracting’s generic realization, and improve greatly the quality of automatic abstracting. Especially, this method uses text clustering, it can not only improve greatly the quality of single document’s automatic abstracting, but also realize multi- document’s automatic abstracting. For a specific plastic domain, an automatic abstracting system TCAAC is implemented.

Key words: Automatic abstract; Text cluster; Natural language understanding

摘要： 针对当前自动文摘方法的不足，提出了基于文本聚类和自然语言理解的自动文摘实现方法。可以克服常规自动文摘方法的不足，使文摘的质量和效果得到大大的提高。将文本聚类引入自动文摘中，不但使单文档的文摘质量得到提高，而且能够实现多文档的自动文摘，这是现有的自动文摘技术所没有涉及的。实现了面向“塑料”行业的基于文本聚类和自然语言理解的自动文摘系统TCAAS。

关键词: 自动文摘；文本聚类；自然语言理解